Artificial Intelligence Nanodegree

Computer Vision Capstone

Project: Facial Keypoint Detection


Welcome to the final Computer Vision project in the Artificial Intelligence Nanodegree program!

In this project, you’ll combine your knowledge of computer vision techniques and deep learning to build and end-to-end facial keypoint recognition system! Facial keypoints include points around the eyes, nose, and mouth on any face and are used in many applications, from facial tracking to emotion recognition.

There are three main parts to this project:

Part 1 : Investigating OpenCV, pre-processing, and face detection

Part 2 : Training a Convolutional Neural Network (CNN) to detect facial keypoints

Part 3 : Putting parts 1 and 2 together to identify facial keypoints on any image!


*Here's what you need to know to complete the project:

  1. In this notebook, some template code has already been provided for you, and you will need to implement additional functionality to successfully complete this project. You will not need to modify the included code beyond what is requested.

    a. Sections that begin with '(IMPLEMENTATION)' in the header indicate that the following block of code will require additional functionality which you must provide. Instructions will be provided for each section, and the specifics of the implementation are marked in the code block with a 'TODO' statement. Please be sure to read the instructions carefully!

  1. In addition to implementing code, there will be questions that you must answer which relate to the project and your implementation.

    a. Each section where you will answer a question is preceded by a 'Question X' header.

    b. Carefully read each question and provide thorough answers in the following text boxes that begin with 'Answer:'.

Note: Code and Markdown cells can be executed using the Shift + Enter keyboard shortcut. Markdown cells can be edited by double-clicking the cell to enter edit mode.

The rubric contains optional suggestions for enhancing the project beyond the minimum requirements. If you decide to pursue the "(Optional)" sections, you should include the code in this IPython notebook.

Your project submission will be evaluated based on your answers to each of the questions and the code implementations you provide.

Steps to Complete the Project

Each part of the notebook is further broken down into separate steps. Feel free to use the links below to navigate the notebook.

In this project you will get to explore a few of the many computer vision algorithms built into the OpenCV library. This expansive computer vision library is now almost 20 years old and still growing!

The project itself is broken down into three large parts, then even further into separate steps. Make sure to read through each step, and complete any sections that begin with '(IMPLEMENTATION)' in the header; these implementation sections may contain multiple TODOs that will be marked in code. For convenience, we provide links to each of these steps below.

Part 1 : Investigating OpenCV, pre-processing, and face detection

  • Step 0: Detect Faces Using a Haar Cascade Classifier
  • Step 1: Add Eye Detection
  • Step 2: De-noise an Image for Better Face Detection
  • Step 3: Blur an Image and Perform Edge Detection
  • Step 4: Automatically Hide the Identity of an Individual

Part 2 : Training a Convolutional Neural Network (CNN) to detect facial keypoints

  • Step 5: Create a CNN to Recognize Facial Keypoints
  • Step 6: Compile and Train the Model
  • Step 7: Visualize the Loss and Answer Questions

Part 3 : Putting parts 1 and 2 together to identify facial keypoints on any image!

  • Step 8: Build a Robust Facial Keypoints Detector (Complete the CV Pipeline)

Step 0: Detect Faces Using a Haar Cascade Classifier

Have you ever wondered how Facebook automatically tags images with your friends' faces? Or how high-end cameras automatically find and focus on a certain person's face? Applications like these depend heavily on the machine learning task known as face detection - which is the task of automatically finding faces in images containing people.

At its root face detection is a classification problem - that is a problem of distinguishing between distinct classes of things. With face detection these distinct classes are 1) images of human faces and 2) everything else.

We use OpenCV's implementation of Haar feature-based cascade classifiers to detect human faces in images. OpenCV provides many pre-trained face detectors, stored as XML files on github. We have downloaded one of these detectors and stored it in the detector_architectures directory.

Import Resources

In the next python cell, we load in the required libraries for this section of the project.

In [1]:
# Import required libraries for this section

%matplotlib inline

import numpy as np
import matplotlib.pyplot as plt
import math
import cv2                     # OpenCV library for computer vision
from PIL import Image
import time 

Next, we load in and display a test image for performing face detection.

Note: by default OpenCV assumes the ordering of our image's color channels are Blue, then Green, then Red. This is slightly out of order with most image types we'll use in these experiments, whose color channels are ordered Red, then Green, then Blue. In order to switch the Blue and Red channels of our test image around we will use OpenCV's cvtColor function, which you can read more about by checking out some of its documentation located here. This is a general utility function that can do other transformations too like converting a color image to grayscale, and transforming a standard color image to HSV color space.

In [2]:
# Load in color image for face detection
image = cv2.imread('images/test_image_1.jpg')

# Convert the image to RGB colorspace
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)

# Plot our image using subplots to specify a size and title
fig = plt.figure(figsize = (8,8))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Original Image')
ax1.imshow(image)
Out[2]:
<matplotlib.image.AxesImage at 0x7f314920f860>

There are a lot of people - and faces - in this picture. 13 faces to be exact! In the next code cell, we demonstrate how to use a Haar Cascade classifier to detect all the faces in this test image.

This face detector uses information about patterns of intensity in an image to reliably detect faces under varying light conditions. So, to use this face detector, we'll first convert the image from color to grayscale.

Then, we load in the fully trained architecture of the face detector -- found in the file haarcascade_frontalface_default.xml - and use it on our image to find faces!

To learn more about the parameters of the detector see this post.

In [3]:
# Convert the RGB  image to grayscale
gray = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)

# Extract the pre-trained face detector from an xml file
face_cascade = cv2.CascadeClassifier('detector_architectures/haarcascade_frontalface_default.xml')

# Detect the faces in image
faces = face_cascade.detectMultiScale(gray, 4, 6)

# Print the number of faces detected in the image
print('Number of faces detected:', len(faces))

# Make a copy of the orginal image to draw face detections on
image_with_detections = np.copy(image)

# Get the bounding box for each detected face
for (x,y,w,h) in faces:
    # Add a red bounding box to the detections image
    cv2.rectangle(image_with_detections, (x,y), (x+w,y+h), (255,0,0), 3)
    

# Display the image with the detections
fig = plt.figure(figsize = (8,8))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Image with Face Detections')
ax1.imshow(image_with_detections)
Number of faces detected: 13
Out[3]:
<matplotlib.image.AxesImage at 0x7f3148997a58>

In the above code, faces is a numpy array of detected faces, where each row corresponds to a detected face. Each detected face is a 1D array with four entries that specifies the bounding box of the detected face. The first two entries in the array (extracted in the above code as x and y) specify the horizontal and vertical positions of the top left corner of the bounding box. The last two entries in the array (extracted here as w and h) specify the width and height of the box.


Step 1: Add Eye Detections

There are other pre-trained detectors available that use a Haar Cascade Classifier - including full human body detectors, license plate detectors, and more. A full list of the pre-trained architectures can be found here.

To test your eye detector, we'll first read in a new test image with just a single face.

In [4]:
# Load in color image for face detection
image = cv2.imread('images/james.jpg')

# Convert the image to RGB colorspace
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)

# Plot the RGB image
fig = plt.figure(figsize = (6,6))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Original Image')
ax1.imshow(image)
Out[4]:
<matplotlib.image.AxesImage at 0x7f31445332b0>

Notice that even though the image is a black and white image, we have read it in as a color image and so it will still need to be converted to grayscale in order to perform the most accurate face detection.

So, the next steps will be to convert this image to grayscale, then load OpenCV's face detector and run it with parameters that detect this face accurately.

In [5]:
# Convert the RGB  image to grayscale
gray = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)

# Extract the pre-trained face detector from an xml file
face_cascade = cv2.CascadeClassifier('detector_architectures/haarcascade_frontalface_default.xml')

# Detect the faces in image
faces = face_cascade.detectMultiScale(gray, 1.25, 6)

# Print the number of faces detected in the image
print('Number of faces detected:', len(faces))

# Make a copy of the orginal image to draw face detections on
image_with_detections = np.copy(image)

# Get the bounding box for each detected face
for (x,y,w,h) in faces:
    # Add a red bounding box to the detections image
    cv2.rectangle(image_with_detections, (x,y), (x+w,y+h), (255,0,0), 3)
    

# Display the image with the detections
fig = plt.figure(figsize = (6,6))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Image with Face Detection')
ax1.imshow(image_with_detections)
Number of faces detected: 1
Out[5]:
<matplotlib.image.AxesImage at 0x7f314448f240>

(IMPLEMENTATION) Add an eye detector to the current face detection setup.

A Haar-cascade eye detector can be included in the same way that the face detector was and, in this first task, it will be your job to do just this.

To set up an eye detector, use the stored parameters of the eye cascade detector, called haarcascade_eye.xml, located in the detector_architectures subdirectory. In the next code cell, create your eye detector and store its detections.

A few notes before you get started:

First, make sure to give your loaded eye detector the variable name

eye_cascade

and give the list of eye regions you detect the variable name

eyes

Second, since we've already run the face detector over this image, you should only search for eyes within the rectangular face regions detected in faces. This will minimize false detections.

Lastly, once you've run your eye detector over the facial detection region, you should display the RGB image with both the face detection boxes (in red) and your eye detections (in green) to verify that everything works as expected.

In [6]:
# Make a copy of the original image to plot rectangle detections
image_with_detections = np.copy(image)   

# Loop over the detections and draw their corresponding face detection boxes
for (x,y,w,h) in faces:
    cv2.rectangle(image_with_detections, (x,y), (x+w,y+h),(255,0,0), 3)  
    
# Do not change the code above this comment!

    
## TODO: Add eye detection, using haarcascade_eye.xml, to the current face detector algorithm
eye_cascade = cv2.CascadeClassifier('detector_architectures/haarcascade_eye.xml')
eyes = []
for (x,y,w,h) in faces:
    eyes += [(x+ex,y+ey,ew,eh) for (ex,ey,ew,eh) in eye_cascade.detectMultiScale(gray[y:y+h, x:x+w], 1.05, 4)]
## TODO: Loop over the eye detections and draw their corresponding boxes in green on image_with_detections
for (x,y,w,h) in eyes:
    cv2.rectangle(image_with_detections, (x,y), (x+w,y+h),(0,255,0), 3)  

# Plot the image with both faces and eyes detected
fig = plt.figure(figsize = (6,6))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Image with Face and Eye Detection')
ax1.imshow(image_with_detections)
Out[6]:
<matplotlib.image.AxesImage at 0x7f31444624a8>

(Optional) Add face and eye detection to your laptop camera

It's time to kick it up a notch, and add face and eye detection to your laptop's camera! Afterwards, you'll be able to show off your creation like in the gif shown below - made with a completed version of the code!

Notice that not all of the detections here are perfect - and your result need not be perfect either. You should spend a small amount of time tuning the parameters of your detectors to get reasonable results, but don't hold out for perfection. If we wanted perfection we'd need to spend a ton of time tuning the parameters of each detector, cleaning up the input image frames, etc. You can think of this as more of a rapid prototype.

The next cell contains code for a wrapper function called laptop_camera_face_eye_detector that, when called, will activate your laptop's camera. You will place the relevant face and eye detection code in this wrapper function to implement face/eye detection and mark those detections on each image frame that your camera captures.

Before adding anything to the function, you can run it to get an idea of how it works - a small window should pop up showing you the live feed from your camera; you can press any key to close this window.

Note: Mac users may find that activating this function kills the kernel of their notebook every once in a while. If this happens to you, just restart your notebook's kernel, activate cell(s) containing any crucial import statements, and you'll be good to go!

In [17]:
### Add face and eye detection to this laptop camera function 
# Make sure to draw out all faces/eyes found in each frame on the shown video feed

import cv2
import time 

# wrapper function for face/eye detection with your laptop camera
def laptop_camera_go():
    # Create instance of video capturer
    cv2.namedWindow("face detection activated")
    vc = cv2.VideoCapture(0)

    # Try to get the first frame
    if vc.isOpened(): 
        rval, frame = vc.read()
    else:
        rval = False
    
    # Keep the video stream open
    while rval:
        # Plot the image from camera with all the face and eye detections marked
        cv2.imshow("face detection activated", frame)
        
        # Exit functionality - press any key to exit laptop video
        key = cv2.waitKey(20)
        if key > 0: # Exit by pressing any key
            # Destroy windows 
            cv2.destroyAllWindows()
            
            # Make sure window closes on OSx
            for i in range (1,5):
                cv2.waitKey(1)
            return
        
        # Read next frame
        time.sleep(0.05)             # control framerate for computation - default 20 frames per sec
        rval, frame = vc.read()    
In [ ]:
# Call the laptop camera face/eye detector function above
laptop_camera_go()

Step 2: De-noise an Image for Better Face Detection

Image quality is an important aspect of any computer vision task. Typically, when creating a set of images to train a deep learning network, significant care is taken to ensure that training images are free of visual noise or artifacts that hinder object detection. While computer vision algorithms - like a face detector - are typically trained on 'nice' data such as this, new test data doesn't always look so nice!

When applying a trained computer vision algorithm to a new piece of test data one often cleans it up first before feeding it in. This sort of cleaning - referred to as pre-processing - can include a number of cleaning phases like blurring, de-noising, color transformations, etc., and many of these tasks can be accomplished using OpenCV.

In this short subsection we explore OpenCV's noise-removal functionality to see how we can clean up a noisy image, which we then feed into our trained face detector.

Create a noisy image to work with

In the next cell, we create an artificial noisy version of the previous multi-face image. This is a little exaggerated - we don't typically get images that are this noisy - but image noise, or 'grainy-ness' in a digitial image - is a fairly common phenomenon.

In [7]:
# Load in the multi-face test image again
image = cv2.imread('images/test_image_1.jpg')

# Convert the image copy to RGB colorspace
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)

# Make an array copy of this image
image_with_noise = np.asarray(image)

# Create noise - here we add noise sampled randomly from a Gaussian distribution: a common model for noise
noise_level = 40
noise = np.random.randn(image.shape[0],image.shape[1],image.shape[2])*noise_level

# Add this noise to the array image copy
image_with_noise = image_with_noise + noise

# Convert back to uint8 format
image_with_noise = np.asarray([np.uint8(np.clip(i,0,255)) for i in image_with_noise])

# Plot our noisy image!
fig = plt.figure(figsize = (8,8))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Noisy Image')
ax1.imshow(image_with_noise)
Out[7]:
<matplotlib.image.AxesImage at 0x7f3144434ac8>

In the context of face detection, the problem with an image like this is that - due to noise - we may miss some faces or get false detections.

In the next cell we apply the same trained OpenCV detector with the same settings as before, to see what sort of detections we get.

In [8]:
# Convert the RGB  image to grayscale
gray_noise = cv2.cvtColor(image_with_noise, cv2.COLOR_RGB2GRAY)

# Extract the pre-trained face detector from an xml file
face_cascade = cv2.CascadeClassifier('detector_architectures/haarcascade_frontalface_default.xml')

# Detect the faces in image
faces = face_cascade.detectMultiScale(gray_noise, 4, 6)

# Print the number of faces detected in the image
print('Number of faces detected:', len(faces))

# Make a copy of the orginal image to draw face detections on
image_with_detections = np.copy(image_with_noise)

# Get the bounding box for each detected face
for (x,y,w,h) in faces:
    # Add a red bounding box to the detections image
    cv2.rectangle(image_with_detections, (x,y), (x+w,y+h), (255,0,0), 3)
    

# Display the image with the detections
fig = plt.figure(figsize = (8,8))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Noisy Image with Face Detections')
ax1.imshow(image_with_detections)
Number of faces detected: 13
Out[8]:
<matplotlib.image.AxesImage at 0x7f314437deb8>

With this added noise we now miss one of the faces!

(IMPLEMENTATION) De-noise this image for better face detection

Time to get your hands dirty: using OpenCV's built in color image de-noising functionality called fastNlMeansDenoisingColored - de-noise this image enough so that all the faces in the image are properly detected. Once you have cleaned the image in the next cell, use the cell that follows to run our trained face detector over the cleaned image to check out its detections.

You can find its official documentation here and a useful example here.

Note: you can keep all parameters except photo_render fixed as shown in the second link above. Play around with the value of this parameter - see how it affects the resulting cleaned image.

In [11]:
## TODO: Use OpenCV's built in color image de-noising function to clean up our noisy image!


denoised_image = cv2.fastNlMeansDenoisingColored(np.copy(image_with_noise), None, 8, 8, 7, 21) # your final de-noised image (should be RGB)
In [12]:
## TODO: Run the face detector on the de-noised image to improve your detections and display the result
# Convert the RGB image to grayscale
gray_denoise = cv2.cvtColor(denoised_image, cv2.COLOR_RGB2GRAY)

# Detect the faces in image
faces = face_cascade.detectMultiScale(gray_denoise, 4, 6)

# Print the number of faces detected in the image
print('Number of faces detected:', len(faces))

# Make a copy of the orginal image to draw face detections on
image_with_detections = np.copy(denoised_image)

# Get the bounding box for each detected face
for (x,y,w,h) in faces:
    # Add a red bounding box to the detections image
    cv2.rectangle(image_with_detections, (x,y), (x+w,y+h), (255,0,0), 3)
    

# Display the image with the detections
fig = plt.figure(figsize = (8,8))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Noisy Image with Face Detections')
ax1.imshow(image_with_detections)
Number of faces detected: 13
Out[12]:
<matplotlib.image.AxesImage at 0x7f314432f7f0>

Step 3: Blur an Image and Perform Edge Detection

Now that we have developed a simple pipeline for detecting faces using OpenCV - let's start playing around with a few fun things we can do with all those detected faces!

Importance of Blur in Edge Detection

Edge detection is a concept that pops up almost everywhere in computer vision applications, as edge-based features (as well as features built on top of edges) are often some of the best features for e.g., object detection and recognition problems.

Edge detection is a dimension reduction technique - by keeping only the edges of an image we get to throw away a lot of non-discriminating information. And typically the most useful kind of edge-detection is one that preserves only the important, global structures (ignoring local structures that aren't very discriminative). So removing local structures / retaining global structures is a crucial pre-processing step to performing edge detection in an image, and blurring can do just that.

Below is an animated gif showing the result of an edge-detected cat taken from Wikipedia, where the image is gradually blurred more and more prior to edge detection. When the animation begins you can't quite make out what it's a picture of, but as the animation evolves and local structures are removed via blurring the cat becomes visible in the edge-detected image.

Edge detection is a convolution performed on the image itself, and you can read about Canny edge detection on this OpenCV documentation page.

Canny edge detection

In the cell below we load in a test image, then apply Canny edge detection on it. The original image is shown on the left panel of the figure, while the edge-detected version of the image is shown on the right. Notice how the result looks very busy - there are too many little details preserved in the image before it is sent to the edge detector. When applied in computer vision applications, edge detection should preserve global structure; doing away with local structures that don't help describe what objects are in the image.

In [13]:
# Load in the image
image = cv2.imread('images/fawzia.jpg')

# Convert to RGB colorspace
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)

# Convert to grayscale
gray = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)  

# Perform Canny edge detection
edges = cv2.Canny(gray,100,200)

# Dilate the image to amplify edges
edges = cv2.dilate(edges, None)

# Plot the RGB and edge-detected image
fig = plt.figure(figsize = (15,15))
ax1 = fig.add_subplot(121)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Original Image')
ax1.imshow(image)

ax2 = fig.add_subplot(122)
ax2.set_xticks([])
ax2.set_yticks([])

ax2.set_title('Canny Edges')
ax2.imshow(edges, cmap='gray')
Out[13]:
<matplotlib.image.AxesImage at 0x7f314429e748>

Without first blurring the image, and removing small, local structures, a lot of irrelevant edge content gets picked up and amplified by the detector (as shown in the right panel above).

(IMPLEMENTATION) Blur the image then perform edge detection

In the next cell, you will repeat this experiment - blurring the image first to remove these local structures, so that only the important boudnary details remain in the edge-detected image.

Blur the image by using OpenCV's filter2d functionality - which is discussed in this documentation page - and use an averaging kernel of width equal to 4.

In [14]:
### TODO: Blur the test image using OpenCV's filter2d functionality, 
# Use an averaging kernel, and a kernel width equal to 4
kernel = np.ones((4,4),np.float32)/16
blur_img = cv2.filter2D(image,-1,kernel)

## TODO: Then perform Canny edge detection and display the output
# Convert to grayscale
gray = cv2.cvtColor(blur_img, cv2.COLOR_RGB2GRAY)  

# Perform Canny edge detection
edges = cv2.Canny(gray,100,200)

# Dilate the image to amplify edges
edges = cv2.dilate(edges, None)

# Plot the RGB and edge-detected image
fig = plt.figure(figsize = (15,15))
ax1 = fig.add_subplot(121)
ax1.set_xticks([])
ax1.set_yticks([])

ax1.set_title('Blurred Image')
ax1.imshow(blur_img)

ax2 = fig.add_subplot(122)
ax2.set_xticks([])
ax2.set_yticks([])

ax2.set_title('Canny Edges')
ax2.imshow(edges, cmap='gray')
Out[14]:
<matplotlib.image.AxesImage at 0x7f314422d8d0>

Step 4: Automatically Hide the Identity of an Individual

If you film something like a documentary or reality TV, you must get permission from every individual shown on film before you can show their face, otherwise you need to blur it out - by blurring the face a lot (so much so that even the global structures are obscured)! This is also true for projects like Google's StreetView maps - an enormous collection of mapping images taken from a fleet of Google vehicles. Because it would be impossible for Google to get the permission of every single person accidentally captured in one of these images they blur out everyone's faces, the detected images must automatically blur the identity of detected people. Here's a few examples of folks caught in the camera of a Google street view vehicle.

Read in an image to perform identity detection

Let's try this out for ourselves. Use the face detection pipeline built above and what you know about using the filter2D to blur and image, and use these in tandem to hide the identity of the person in the following image - loaded in and printed in the next cell.

In [15]:
# Load in the image
image = cv2.imread('images/gus.jpg')

# Convert the image to RGB colorspace
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)

# Display the image
fig = plt.figure(figsize = (6,6))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])
ax1.set_title('Original Image')
ax1.imshow(image)
Out[15]:
<matplotlib.image.AxesImage at 0x7f3144145e10>

(IMPLEMENTATION) Use blurring to hide the identity of an individual in an image

The idea here is to 1) automatically detect the face in this image, and then 2) blur it out! Make sure to adjust the parameters of the averaging blur filter to completely obscure this person's identity.

In [18]:
## TODO: Implement face detection
gray = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)
faces = face_cascade.detectMultiScale(gray, 1.23, 6)

## TODO: Blur the bounding box around each detected face using an averaging filter and display the result
anonymized_image = np.copy(image)
for (x,y,w,h) in faces:
    anonymized_image[y:y+h, x:x+w] = cv2.blur(anonymized_image[y:y+h, x:x+w], (100,100))

fig = plt.figure(figsize = (6,6))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])
ax1.set_title('Anonymized Image')
ax1.imshow(anonymized_image)
Out[18]:
<matplotlib.image.AxesImage at 0x7f3144046198>

(Optional) Build identity protection into your laptop camera

In this optional task you can add identity protection to your laptop camera, using the previously completed code where you added face detection to your laptop camera - and the task above. You should be able to get reasonable results with little parameter tuning - like the one shown in the gif below.

As with the previous video task, to make this perfect would require significant effort - so don't strive for perfection here, strive for reasonable quality.

The next cell contains code a wrapper function called laptop_camera_identity_hider that - when called - will activate your laptop's camera. You need to place the relevant face detection and blurring code developed above in this function in order to blur faces entering your laptop camera's field of view.

Before adding anything to the function you can call it to get a hang of how it works - a small window will pop up showing you the live feed from your camera, you can press any key to close this window.

Note: Mac users may find that activating this function kills the kernel of their notebook every once in a while. If this happens to you, just restart your notebook's kernel, activate cell(s) containing any crucial import statements, and you'll be good to go!

In [ ]:
### Insert face detection and blurring code into the wrapper below to create an identity protector on your laptop!
import cv2
import time 

def laptop_camera_go():
    # Create instance of video capturer
    cv2.namedWindow("face detection activated")
    vc = cv2.VideoCapture(0)

    # Try to get the first frame
    if vc.isOpened(): 
        rval, frame = vc.read()
    else:
        rval = False
    
    # Keep video stream open
    while rval:
        # Plot image from camera with detections marked
        cv2.imshow("face detection activated", frame)
        
        # Exit functionality - press any key to exit laptop video
        key = cv2.waitKey(20)
        if key > 0: # Exit by pressing any key
            # Destroy windows
            cv2.destroyAllWindows()
            
            for i in range (1,5):
                cv2.waitKey(1)
            return
        
        # Read next frame
        time.sleep(0.05)             # control framerate for computation - default 20 frames per sec
        rval, frame = vc.read()    
        
In [ ]:
# Run laptop identity hider
laptop_camera_go()

Step 5: Create a CNN to Recognize Facial Keypoints

OpenCV is often used in practice with other machine learning and deep learning libraries to produce interesting results. In this stage of the project you will create your own end-to-end pipeline - employing convolutional networks in keras along with OpenCV - to apply a "selfie" filter to streaming video and images.

You will start by creating and then training a convolutional network that can detect facial keypoints in a small dataset of cropped images of human faces. We then guide you towards OpenCV to expanding your detection algorithm to more general images. What are facial keypoints? Let's take a look at some examples.

Facial keypoints (also called facial landmarks) are the small blue-green dots shown on each of the faces in the image above - there are 15 keypoints marked in each image. They mark important areas of the face - the eyes, corners of the mouth, the nose, etc. Facial keypoints can be used in a variety of machine learning applications from face and emotion recognition to commercial applications like the image filters popularized by Snapchat.

Below we illustrate a filter that, using the results of this section, automatically places sunglasses on people in images (using the facial keypoints to place the glasses correctly on each face). Here, the facial keypoints have been colored lime green for visualization purposes.

Make a facial keypoint detector

But first things first: how can we make a facial keypoint detector? Well, at a high level, notice that facial keypoint detection is a regression problem. A single face corresponds to a set of 15 facial keypoints (a set of 15 corresponding $(x, y)$ coordinates, i.e., an output point). Because our input data are images, we can employ a convolutional neural network to recognize patterns in our images and learn how to identify these keypoint given sets of labeled data.

In order to train a regressor, we need a training set - a set of facial image / facial keypoint pairs to train on. For this we will be using this dataset from Kaggle. We've already downloaded this data and placed it in the data directory. Make sure that you have both the training and test data files. The training dataset contains several thousand $96 \times 96$ grayscale images of cropped human faces, along with each face's 15 corresponding facial keypoints (also called landmarks) that have been placed by hand, and recorded in $(x, y)$ coordinates. This wonderful resource also has a substantial testing set, which we will use in tinkering with our convolutional network.

To load in this data, run the Python cell below - notice we will load in both the training and testing sets.

The load_data function is in the included utils.py file.

In [19]:
from utils import *

# Load training set
X_train, y_train = load_data()
print("X_train.shape == {}".format(X_train.shape))
print("y_train.shape == {}; y_train.min == {:.3f}; y_train.max == {:.3f}".format(
    y_train.shape, y_train.min(), y_train.max()))

# Load testing set
X_test, _ = load_data(test=True)
print("X_test.shape == {}".format(X_test.shape))
Using TensorFlow backend.
X_train.shape == (2140, 96, 96, 1)
y_train.shape == (2140, 30); y_train.min == -0.920; y_train.max == 0.996
X_test.shape == (1783, 96, 96, 1)

The load_data function in utils.py originates from this excellent blog post, which you are strongly encouraged to read. Please take the time now to review this function. Note how the output values - that is, the coordinates of each set of facial landmarks - have been normalized to take on values in the range $[-1, 1]$, while the pixel values of each input point (a facial image) have been normalized to the range $[0,1]$.

Note: the original Kaggle dataset contains some images with several missing keypoints. For simplicity, the load_data function removes those images with missing labels from the dataset. As an optional extension, you are welcome to amend the load_data function to include the incomplete data points.

Visualize the Training Data

Execute the code cell below to visualize a subset of the training data.

In [20]:
import matplotlib.pyplot as plt
%matplotlib inline

fig = plt.figure(figsize=(20,20))
fig.subplots_adjust(left=0, right=1, bottom=0, top=1, hspace=0.05, wspace=0.05)
for i in range(9):
    ax = fig.add_subplot(3, 3, i + 1, xticks=[], yticks=[])
    plot_data(X_train[i], y_train[i], ax)

For each training image, there are two landmarks per eyebrow (four total), three per eye (six total), four for the mouth, and one for the tip of the nose.

Review the plot_data function in utils.py to understand how the 30-dimensional training labels in y_train are mapped to facial locations, as this function will prove useful for your pipeline.

(IMPLEMENTATION) Specify the CNN Architecture

In this section, you will specify a neural network for predicting the locations of facial keypoints. Use the code cell below to specify the architecture of your neural network. We have imported some layers that you may find useful for this task, but if you need to use more Keras layers, feel free to import them in the cell.

Your network should accept a $96 \times 96$ grayscale image as input, and it should output a vector with 30 entries, corresponding to the predicted (horizontal and vertical) locations of 15 facial keypoints. If you are not sure where to start, you can find some useful starting architectures in this blog, but you are not permitted to copy any of the architectures that you find online.

In [43]:
# Import deep learning resources from Keras
from keras.models import Sequential
from keras.layers import Convolution2D, MaxPooling2D, Dropout
from keras.layers import Flatten, Dense


## TODO: Specify a CNN architecture
# Your model should accept 96x96 pixel graysale images in
# It should have a fully-connected output layer with 30 values (2 for each facial keypoint)

model = Sequential()
model.add(Convolution2D(32, 3, padding='same', activation='relu', input_shape=X_train.shape[1:]))
model.add(Convolution2D(32, 3, padding='same', activation='relu'))
model.add(MaxPooling2D(pool_size=2))
model.add(Dropout(0.25))

model.add(Convolution2D(64, 3, padding='same', activation='relu'))
model.add(MaxPooling2D(pool_size=3))
model.add(Convolution2D(128, 2, padding='same', activation='relu'))
model.add(MaxPooling2D(pool_size=2))
model.add(Convolution2D(256, 2, padding='same', activation='relu'))
model.add(MaxPooling2D(pool_size=2))
model.add(Dropout(0.25))

model.add(Flatten())
model.add(Dense(512, activation='relu'))
model.add(Dense(512, activation='relu'))
model.add(Dropout(0.25))
model.add(Dense(30, activation='tanh'))

# Summarize the model
model.summary()
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
conv2d_47 (Conv2D)           (None, 96, 96, 32)        320       
_________________________________________________________________
conv2d_48 (Conv2D)           (None, 96, 96, 32)        9248      
_________________________________________________________________
max_pooling2d_37 (MaxPooling (None, 48, 48, 32)        0         
_________________________________________________________________
dropout_31 (Dropout)         (None, 48, 48, 32)        0         
_________________________________________________________________
conv2d_49 (Conv2D)           (None, 48, 48, 64)        18496     
_________________________________________________________________
max_pooling2d_38 (MaxPooling (None, 16, 16, 64)        0         
_________________________________________________________________
conv2d_50 (Conv2D)           (None, 16, 16, 128)       32896     
_________________________________________________________________
max_pooling2d_39 (MaxPooling (None, 8, 8, 128)         0         
_________________________________________________________________
conv2d_51 (Conv2D)           (None, 8, 8, 256)         131328    
_________________________________________________________________
max_pooling2d_40 (MaxPooling (None, 4, 4, 256)         0         
_________________________________________________________________
dropout_32 (Dropout)         (None, 4, 4, 256)         0         
_________________________________________________________________
flatten_11 (Flatten)         (None, 4096)              0         
_________________________________________________________________
dense_31 (Dense)             (None, 512)               2097664   
_________________________________________________________________
dense_32 (Dense)             (None, 512)               262656    
_________________________________________________________________
dropout_33 (Dropout)         (None, 512)               0         
_________________________________________________________________
dense_33 (Dense)             (None, 30)                15390     
=================================================================
Total params: 2,567,998
Trainable params: 2,567,998
Non-trainable params: 0
_________________________________________________________________

Step 6: Compile and Train the Model

After specifying your architecture, you'll need to compile and train the model to detect facial keypoints'

(IMPLEMENTATION) Compile and Train the Model

Use the compile method to configure the learning process. Experiment with your choice of optimizer; you may have some ideas about which will work best (SGD vs. RMSprop, etc), but take the time to empirically verify your theories.

Use the fit method to train the model. Break off a validation set by setting validation_split=0.2. Save the returned History object in the history variable.

Experiment with your model to minimize the validation loss (measured as mean squared error). A very good model will achieve about 0.0015 loss (though it's possible to do even better). When you have finished training, save your model as an HDF5 file with file path my_model.h5.

In [39]:
from keras.optimizers import SGD, RMSprop, Adagrad, Adadelta, Adam, Adamax, Nadam

## TODO: Compile the model
opt = RMSprop(lr=0.0001, decay=1e-6)
model.compile(loss='mean_squared_error', optimizer=opt)

## TODO: Train the model
hist_RMSprop = model.fit(X_train, y_train, batch_size=64, epochs=100, verbose=2, validation_split=0.2)

## TODO: Save the model as model.h5
model.save('my_model_RMSprop.h5')
Train on 1712 samples, validate on 428 samples
Epoch 1/100
3s - loss: 0.0338 - val_loss: 0.0321
Epoch 2/100
3s - loss: 0.0151 - val_loss: 0.0373
Epoch 3/100
3s - loss: 0.0128 - val_loss: 0.0301
Epoch 4/100
3s - loss: 0.0115 - val_loss: 0.0142
Epoch 5/100
3s - loss: 0.0104 - val_loss: 0.0227
Epoch 6/100
3s - loss: 0.0093 - val_loss: 0.0197
Epoch 7/100
3s - loss: 0.0086 - val_loss: 0.0090
Epoch 8/100
3s - loss: 0.0084 - val_loss: 0.0094
Epoch 9/100
3s - loss: 0.0077 - val_loss: 0.0073
Epoch 10/100
3s - loss: 0.0074 - val_loss: 0.0056
Epoch 11/100
3s - loss: 0.0073 - val_loss: 0.0090
Epoch 12/100
3s - loss: 0.0070 - val_loss: 0.0077
Epoch 13/100
3s - loss: 0.0068 - val_loss: 0.0067
Epoch 14/100
3s - loss: 0.0065 - val_loss: 0.0068
Epoch 15/100
3s - loss: 0.0063 - val_loss: 0.0071
Epoch 16/100
3s - loss: 0.0062 - val_loss: 0.0069
Epoch 17/100
3s - loss: 0.0060 - val_loss: 0.0041
Epoch 18/100
3s - loss: 0.0059 - val_loss: 0.0043
Epoch 19/100
3s - loss: 0.0058 - val_loss: 0.0048
Epoch 20/100
3s - loss: 0.0055 - val_loss: 0.0046
Epoch 21/100
3s - loss: 0.0054 - val_loss: 0.0039
Epoch 22/100
3s - loss: 0.0052 - val_loss: 0.0037
Epoch 23/100
3s - loss: 0.0052 - val_loss: 0.0036
Epoch 24/100
3s - loss: 0.0049 - val_loss: 0.0034
Epoch 25/100
3s - loss: 0.0049 - val_loss: 0.0033
Epoch 26/100
3s - loss: 0.0047 - val_loss: 0.0045
Epoch 27/100
3s - loss: 0.0045 - val_loss: 0.0031
Epoch 28/100
3s - loss: 0.0044 - val_loss: 0.0030
Epoch 29/100
3s - loss: 0.0043 - val_loss: 0.0040
Epoch 30/100
3s - loss: 0.0042 - val_loss: 0.0049
Epoch 31/100
3s - loss: 0.0040 - val_loss: 0.0027
Epoch 32/100
3s - loss: 0.0040 - val_loss: 0.0028
Epoch 33/100
3s - loss: 0.0037 - val_loss: 0.0026
Epoch 34/100
3s - loss: 0.0037 - val_loss: 0.0025
Epoch 35/100
3s - loss: 0.0035 - val_loss: 0.0025
Epoch 36/100
3s - loss: 0.0035 - val_loss: 0.0024
Epoch 37/100
3s - loss: 0.0035 - val_loss: 0.0024
Epoch 38/100
3s - loss: 0.0033 - val_loss: 0.0024
Epoch 39/100
3s - loss: 0.0033 - val_loss: 0.0022
Epoch 40/100
3s - loss: 0.0032 - val_loss: 0.0030
Epoch 41/100
3s - loss: 0.0031 - val_loss: 0.0022
Epoch 42/100
3s - loss: 0.0030 - val_loss: 0.0027
Epoch 43/100
3s - loss: 0.0030 - val_loss: 0.0026
Epoch 44/100
3s - loss: 0.0030 - val_loss: 0.0022
Epoch 45/100
3s - loss: 0.0029 - val_loss: 0.0027
Epoch 46/100
3s - loss: 0.0029 - val_loss: 0.0021
Epoch 47/100
3s - loss: 0.0028 - val_loss: 0.0020
Epoch 48/100
3s - loss: 0.0027 - val_loss: 0.0022
Epoch 49/100
3s - loss: 0.0027 - val_loss: 0.0019
Epoch 50/100
3s - loss: 0.0026 - val_loss: 0.0017
Epoch 51/100
3s - loss: 0.0027 - val_loss: 0.0018
Epoch 52/100
3s - loss: 0.0025 - val_loss: 0.0018
Epoch 53/100
3s - loss: 0.0025 - val_loss: 0.0016
Epoch 54/100
3s - loss: 0.0024 - val_loss: 0.0016
Epoch 55/100
3s - loss: 0.0024 - val_loss: 0.0016
Epoch 56/100
3s - loss: 0.0024 - val_loss: 0.0019
Epoch 57/100
3s - loss: 0.0023 - val_loss: 0.0016
Epoch 58/100
3s - loss: 0.0023 - val_loss: 0.0022
Epoch 59/100
3s - loss: 0.0023 - val_loss: 0.0018
Epoch 60/100
3s - loss: 0.0022 - val_loss: 0.0015
Epoch 61/100
3s - loss: 0.0022 - val_loss: 0.0015
Epoch 62/100
3s - loss: 0.0021 - val_loss: 0.0015
Epoch 63/100
3s - loss: 0.0021 - val_loss: 0.0014
Epoch 64/100
3s - loss: 0.0021 - val_loss: 0.0016
Epoch 65/100
3s - loss: 0.0021 - val_loss: 0.0014
Epoch 66/100
3s - loss: 0.0020 - val_loss: 0.0014
Epoch 67/100
3s - loss: 0.0020 - val_loss: 0.0016
Epoch 68/100
3s - loss: 0.0020 - val_loss: 0.0013
Epoch 69/100
3s - loss: 0.0019 - val_loss: 0.0013
Epoch 70/100
3s - loss: 0.0019 - val_loss: 0.0013
Epoch 71/100
3s - loss: 0.0019 - val_loss: 0.0013
Epoch 72/100
3s - loss: 0.0018 - val_loss: 0.0018
Epoch 73/100
3s - loss: 0.0018 - val_loss: 0.0012
Epoch 74/100
3s - loss: 0.0019 - val_loss: 0.0014
Epoch 75/100
3s - loss: 0.0018 - val_loss: 0.0012
Epoch 76/100
3s - loss: 0.0018 - val_loss: 0.0018
Epoch 77/100
3s - loss: 0.0018 - val_loss: 0.0014
Epoch 78/100
3s - loss: 0.0017 - val_loss: 0.0012
Epoch 79/100
3s - loss: 0.0018 - val_loss: 0.0012
Epoch 80/100
3s - loss: 0.0016 - val_loss: 0.0021
Epoch 81/100
3s - loss: 0.0017 - val_loss: 0.0012
Epoch 82/100
3s - loss: 0.0016 - val_loss: 0.0011
Epoch 83/100
3s - loss: 0.0016 - val_loss: 0.0012
Epoch 84/100
3s - loss: 0.0016 - val_loss: 0.0012
Epoch 85/100
3s - loss: 0.0016 - val_loss: 0.0015
Epoch 86/100
3s - loss: 0.0015 - val_loss: 0.0011
Epoch 87/100
3s - loss: 0.0016 - val_loss: 0.0013
Epoch 88/100
3s - loss: 0.0016 - val_loss: 0.0011
Epoch 89/100
3s - loss: 0.0015 - val_loss: 0.0011
Epoch 90/100
3s - loss: 0.0015 - val_loss: 0.0011
Epoch 91/100
3s - loss: 0.0015 - val_loss: 0.0012
Epoch 92/100
3s - loss: 0.0014 - val_loss: 0.0012
Epoch 93/100
3s - loss: 0.0014 - val_loss: 0.0010
Epoch 94/100
3s - loss: 0.0014 - val_loss: 0.0011
Epoch 95/100
3s - loss: 0.0014 - val_loss: 0.0011
Epoch 96/100
3s - loss: 0.0014 - val_loss: 0.0012
Epoch 97/100
3s - loss: 0.0013 - val_loss: 0.0012
Epoch 98/100
3s - loss: 0.0014 - val_loss: 0.0010
Epoch 99/100
3s - loss: 0.0013 - val_loss: 0.0013
Epoch 100/100
3s - loss: 0.0013 - val_loss: 0.0013
In [42]:
## TODO: Compile the model
opt = Adadelta()
model.compile(loss='mean_squared_error', optimizer=opt)

## TODO: Train the model
hist_Adadelta = model.fit(X_train, y_train, batch_size=64, epochs=100, verbose=2, validation_split=0.2)

## TODO: Save the model as model.h5
model.save('my_model_Adadelta.h5')
Train on 1712 samples, validate on 428 samples
Epoch 1/100
3s - loss: 0.0354 - val_loss: 0.0198
Epoch 2/100
3s - loss: 0.0130 - val_loss: 0.0161
Epoch 3/100
3s - loss: 0.0106 - val_loss: 0.0187
Epoch 4/100
3s - loss: 0.0096 - val_loss: 0.0099
Epoch 5/100
3s - loss: 0.0086 - val_loss: 0.0115
Epoch 6/100
3s - loss: 0.0078 - val_loss: 0.0106
Epoch 7/100
3s - loss: 0.0073 - val_loss: 0.0089
Epoch 8/100
3s - loss: 0.0068 - val_loss: 0.0072
Epoch 9/100
3s - loss: 0.0069 - val_loss: 0.0079
Epoch 10/100
3s - loss: 0.0065 - val_loss: 0.0067
Epoch 11/100
3s - loss: 0.0064 - val_loss: 0.0079
Epoch 12/100
3s - loss: 0.0064 - val_loss: 0.0069
Epoch 13/100
3s - loss: 0.0061 - val_loss: 0.0060
Epoch 14/100
3s - loss: 0.0060 - val_loss: 0.0065
Epoch 15/100
3s - loss: 0.0059 - val_loss: 0.0082
Epoch 16/100
3s - loss: 0.0059 - val_loss: 0.0062
Epoch 17/100
3s - loss: 0.0058 - val_loss: 0.0077
Epoch 18/100
3s - loss: 0.0057 - val_loss: 0.0058
Epoch 19/100
3s - loss: 0.0056 - val_loss: 0.0058
Epoch 20/100
3s - loss: 0.0056 - val_loss: 0.0071
Epoch 21/100
3s - loss: 0.0056 - val_loss: 0.0057
Epoch 22/100
3s - loss: 0.0055 - val_loss: 0.0053
Epoch 23/100
3s - loss: 0.0055 - val_loss: 0.0046
Epoch 24/100
3s - loss: 0.0054 - val_loss: 0.0062
Epoch 25/100
3s - loss: 0.0054 - val_loss: 0.0049
Epoch 26/100
3s - loss: 0.0054 - val_loss: 0.0056
Epoch 27/100
3s - loss: 0.0053 - val_loss: 0.0053
Epoch 28/100
3s - loss: 0.0053 - val_loss: 0.0051
Epoch 29/100
3s - loss: 0.0053 - val_loss: 0.0050
Epoch 30/100
3s - loss: 0.0052 - val_loss: 0.0046
Epoch 31/100
3s - loss: 0.0052 - val_loss: 0.0050
Epoch 32/100
3s - loss: 0.0052 - val_loss: 0.0046
Epoch 33/100
3s - loss: 0.0051 - val_loss: 0.0048
Epoch 34/100
3s - loss: 0.0051 - val_loss: 0.0045
Epoch 35/100
3s - loss: 0.0051 - val_loss: 0.0044
Epoch 36/100
3s - loss: 0.0051 - val_loss: 0.0046
Epoch 37/100
3s - loss: 0.0050 - val_loss: 0.0050
Epoch 38/100
3s - loss: 0.0050 - val_loss: 0.0050
Epoch 39/100
3s - loss: 0.0050 - val_loss: 0.0045
Epoch 40/100
3s - loss: 0.0050 - val_loss: 0.0048
Epoch 41/100
3s - loss: 0.0050 - val_loss: 0.0048
Epoch 42/100
3s - loss: 0.0050 - val_loss: 0.0047
Epoch 43/100
3s - loss: 0.0049 - val_loss: 0.0047
Epoch 44/100
3s - loss: 0.0049 - val_loss: 0.0051
Epoch 45/100
3s - loss: 0.0049 - val_loss: 0.0044
Epoch 46/100
3s - loss: 0.0048 - val_loss: 0.0043
Epoch 47/100
3s - loss: 0.0048 - val_loss: 0.0043
Epoch 48/100
3s - loss: 0.0048 - val_loss: 0.0045
Epoch 49/100
3s - loss: 0.0048 - val_loss: 0.0045
Epoch 50/100
3s - loss: 0.0048 - val_loss: 0.0045
Epoch 51/100
3s - loss: 0.0047 - val_loss: 0.0045
Epoch 52/100
3s - loss: 0.0048 - val_loss: 0.0043
Epoch 53/100
3s - loss: 0.0048 - val_loss: 0.0042
Epoch 54/100
3s - loss: 0.0047 - val_loss: 0.0043
Epoch 55/100
3s - loss: 0.0047 - val_loss: 0.0042
Epoch 56/100
3s - loss: 0.0047 - val_loss: 0.0046
Epoch 57/100
3s - loss: 0.0047 - val_loss: 0.0042
Epoch 58/100
3s - loss: 0.0046 - val_loss: 0.0044
Epoch 59/100
3s - loss: 0.0046 - val_loss: 0.0042
Epoch 60/100
3s - loss: 0.0046 - val_loss: 0.0042
Epoch 61/100
3s - loss: 0.0046 - val_loss: 0.0042
Epoch 62/100
3s - loss: 0.0046 - val_loss: 0.0041
Epoch 63/100
3s - loss: 0.0046 - val_loss: 0.0045
Epoch 64/100
3s - loss: 0.0046 - val_loss: 0.0041
Epoch 65/100
3s - loss: 0.0045 - val_loss: 0.0042
Epoch 66/100
3s - loss: 0.0045 - val_loss: 0.0042
Epoch 67/100
3s - loss: 0.0045 - val_loss: 0.0043
Epoch 68/100
3s - loss: 0.0045 - val_loss: 0.0041
Epoch 69/100
3s - loss: 0.0045 - val_loss: 0.0042
Epoch 70/100
3s - loss: 0.0045 - val_loss: 0.0043
Epoch 71/100
3s - loss: 0.0044 - val_loss: 0.0040
Epoch 72/100
3s - loss: 0.0044 - val_loss: 0.0040
Epoch 73/100
3s - loss: 0.0044 - val_loss: 0.0040
Epoch 74/100
3s - loss: 0.0044 - val_loss: 0.0042
Epoch 75/100
3s - loss: 0.0044 - val_loss: 0.0041
Epoch 76/100
3s - loss: 0.0043 - val_loss: 0.0040
Epoch 77/100
3s - loss: 0.0044 - val_loss: 0.0040
Epoch 78/100
3s - loss: 0.0043 - val_loss: 0.0040
Epoch 79/100
3s - loss: 0.0043 - val_loss: 0.0039
Epoch 80/100
3s - loss: 0.0043 - val_loss: 0.0039
Epoch 81/100
3s - loss: 0.0043 - val_loss: 0.0039
Epoch 82/100
3s - loss: 0.0043 - val_loss: 0.0039
Epoch 83/100
3s - loss: 0.0042 - val_loss: 0.0040
Epoch 84/100
3s - loss: 0.0042 - val_loss: 0.0039
Epoch 85/100
3s - loss: 0.0042 - val_loss: 0.0038
Epoch 86/100
3s - loss: 0.0042 - val_loss: 0.0038
Epoch 87/100
3s - loss: 0.0042 - val_loss: 0.0039
Epoch 88/100
3s - loss: 0.0041 - val_loss: 0.0038
Epoch 89/100
3s - loss: 0.0041 - val_loss: 0.0038
Epoch 90/100
3s - loss: 0.0041 - val_loss: 0.0037
Epoch 91/100
3s - loss: 0.0040 - val_loss: 0.0039
Epoch 92/100
3s - loss: 0.0040 - val_loss: 0.0037
Epoch 93/100
3s - loss: 0.0040 - val_loss: 0.0037
Epoch 94/100
3s - loss: 0.0040 - val_loss: 0.0037
Epoch 95/100
3s - loss: 0.0040 - val_loss: 0.0036
Epoch 96/100
3s - loss: 0.0040 - val_loss: 0.0036
Epoch 97/100
3s - loss: 0.0040 - val_loss: 0.0036
Epoch 98/100
3s - loss: 0.0039 - val_loss: 0.0036
Epoch 99/100
3s - loss: 0.0039 - val_loss: 0.0036
Epoch 100/100
3s - loss: 0.0038 - val_loss: 0.0035
In [44]:
## TODO: Compile the model
opt = SGD(lr=0.01, momentum=0.9, decay=0.0001)
model.compile(loss='mean_squared_error', optimizer=opt)

## TODO: Train the model
hist_SGD = model.fit(X_train, y_train, batch_size=64, epochs=100, verbose=2, validation_split=0.2)

## TODO: Save the model as model.h5
model.save('my_model_SGD.h5')
Train on 1712 samples, validate on 428 samples
Epoch 1/100
3s - loss: 0.1066 - val_loss: 0.0434
Epoch 2/100
2s - loss: 0.0199 - val_loss: 0.0196
Epoch 3/100
2s - loss: 0.0119 - val_loss: 0.0154
Epoch 4/100
2s - loss: 0.0105 - val_loss: 0.0138
Epoch 5/100
2s - loss: 0.0097 - val_loss: 0.0129
Epoch 6/100
2s - loss: 0.0093 - val_loss: 0.0119
Epoch 7/100
2s - loss: 0.0089 - val_loss: 0.0115
Epoch 8/100
2s - loss: 0.0085 - val_loss: 0.0110
Epoch 9/100
2s - loss: 0.0084 - val_loss: 0.0107
Epoch 10/100
2s - loss: 0.0080 - val_loss: 0.0098
Epoch 11/100
2s - loss: 0.0078 - val_loss: 0.0096
Epoch 12/100
3s - loss: 0.0076 - val_loss: 0.0090
Epoch 13/100
2s - loss: 0.0075 - val_loss: 0.0089
Epoch 14/100
3s - loss: 0.0073 - val_loss: 0.0083
Epoch 15/100
2s - loss: 0.0072 - val_loss: 0.0081
Epoch 16/100
3s - loss: 0.0070 - val_loss: 0.0080
Epoch 17/100
2s - loss: 0.0071 - val_loss: 0.0078
Epoch 18/100
2s - loss: 0.0069 - val_loss: 0.0074
Epoch 19/100
2s - loss: 0.0068 - val_loss: 0.0073
Epoch 20/100
2s - loss: 0.0067 - val_loss: 0.0073
Epoch 21/100
2s - loss: 0.0067 - val_loss: 0.0070
Epoch 22/100
2s - loss: 0.0066 - val_loss: 0.0069
Epoch 23/100
2s - loss: 0.0066 - val_loss: 0.0066
Epoch 24/100
2s - loss: 0.0065 - val_loss: 0.0065
Epoch 25/100
2s - loss: 0.0064 - val_loss: 0.0065
Epoch 26/100
3s - loss: 0.0064 - val_loss: 0.0064
Epoch 27/100
3s - loss: 0.0064 - val_loss: 0.0063
Epoch 28/100
3s - loss: 0.0063 - val_loss: 0.0061
Epoch 29/100
2s - loss: 0.0063 - val_loss: 0.0062
Epoch 30/100
3s - loss: 0.0063 - val_loss: 0.0060
Epoch 31/100
2s - loss: 0.0063 - val_loss: 0.0061
Epoch 32/100
2s - loss: 0.0062 - val_loss: 0.0060
Epoch 33/100
2s - loss: 0.0062 - val_loss: 0.0058
Epoch 34/100
3s - loss: 0.0061 - val_loss: 0.0059
Epoch 35/100
3s - loss: 0.0062 - val_loss: 0.0057
Epoch 36/100
2s - loss: 0.0061 - val_loss: 0.0058
Epoch 37/100
3s - loss: 0.0061 - val_loss: 0.0057
Epoch 38/100
2s - loss: 0.0060 - val_loss: 0.0056
Epoch 39/100
2s - loss: 0.0060 - val_loss: 0.0057
Epoch 40/100
2s - loss: 0.0061 - val_loss: 0.0055
Epoch 41/100
3s - loss: 0.0060 - val_loss: 0.0056
Epoch 42/100
3s - loss: 0.0059 - val_loss: 0.0055
Epoch 43/100
3s - loss: 0.0059 - val_loss: 0.0055
Epoch 44/100
2s - loss: 0.0059 - val_loss: 0.0055
Epoch 45/100
2s - loss: 0.0059 - val_loss: 0.0054
Epoch 46/100
3s - loss: 0.0059 - val_loss: 0.0054
Epoch 47/100
2s - loss: 0.0059 - val_loss: 0.0054
Epoch 48/100
3s - loss: 0.0058 - val_loss: 0.0054
Epoch 49/100
2s - loss: 0.0058 - val_loss: 0.0054
Epoch 50/100
2s - loss: 0.0058 - val_loss: 0.0053
Epoch 51/100
3s - loss: 0.0058 - val_loss: 0.0053
Epoch 52/100
2s - loss: 0.0058 - val_loss: 0.0053
Epoch 53/100
3s - loss: 0.0058 - val_loss: 0.0053
Epoch 54/100
2s - loss: 0.0058 - val_loss: 0.0053
Epoch 55/100
2s - loss: 0.0058 - val_loss: 0.0052
Epoch 56/100
3s - loss: 0.0057 - val_loss: 0.0053
Epoch 57/100
3s - loss: 0.0057 - val_loss: 0.0051
Epoch 58/100
2s - loss: 0.0057 - val_loss: 0.0052
Epoch 59/100
2s - loss: 0.0057 - val_loss: 0.0052
Epoch 60/100
2s - loss: 0.0056 - val_loss: 0.0052
Epoch 61/100
3s - loss: 0.0057 - val_loss: 0.0052
Epoch 62/100
3s - loss: 0.0056 - val_loss: 0.0052
Epoch 63/100
3s - loss: 0.0056 - val_loss: 0.0052
Epoch 64/100
3s - loss: 0.0055 - val_loss: 0.0052
Epoch 65/100
2s - loss: 0.0056 - val_loss: 0.0052
Epoch 66/100
2s - loss: 0.0056 - val_loss: 0.0051
Epoch 67/100
2s - loss: 0.0056 - val_loss: 0.0051
Epoch 68/100
2s - loss: 0.0056 - val_loss: 0.0051
Epoch 69/100
2s - loss: 0.0055 - val_loss: 0.0051
Epoch 70/100
2s - loss: 0.0055 - val_loss: 0.0051
Epoch 71/100
3s - loss: 0.0055 - val_loss: 0.0051
Epoch 72/100
3s - loss: 0.0055 - val_loss: 0.0051
Epoch 73/100
3s - loss: 0.0055 - val_loss: 0.0050
Epoch 74/100
2s - loss: 0.0055 - val_loss: 0.0051
Epoch 75/100
2s - loss: 0.0056 - val_loss: 0.0050
Epoch 76/100
2s - loss: 0.0055 - val_loss: 0.0051
Epoch 77/100
3s - loss: 0.0055 - val_loss: 0.0050
Epoch 78/100
3s - loss: 0.0055 - val_loss: 0.0050
Epoch 79/100
2s - loss: 0.0055 - val_loss: 0.0050
Epoch 80/100
3s - loss: 0.0054 - val_loss: 0.0050
Epoch 81/100
2s - loss: 0.0054 - val_loss: 0.0050
Epoch 82/100
3s - loss: 0.0054 - val_loss: 0.0049
Epoch 83/100
2s - loss: 0.0054 - val_loss: 0.0049
Epoch 84/100
2s - loss: 0.0055 - val_loss: 0.0050
Epoch 85/100
3s - loss: 0.0055 - val_loss: 0.0049
Epoch 86/100
3s - loss: 0.0054 - val_loss: 0.0050
Epoch 87/100
3s - loss: 0.0054 - val_loss: 0.0050
Epoch 88/100
2s - loss: 0.0054 - val_loss: 0.0050
Epoch 89/100
2s - loss: 0.0054 - val_loss: 0.0050
Epoch 90/100
2s - loss: 0.0054 - val_loss: 0.0049
Epoch 91/100
3s - loss: 0.0054 - val_loss: 0.0050
Epoch 92/100
3s - loss: 0.0054 - val_loss: 0.0049
Epoch 93/100
3s - loss: 0.0054 - val_loss: 0.0049
Epoch 94/100
2s - loss: 0.0054 - val_loss: 0.0049
Epoch 95/100
2s - loss: 0.0053 - val_loss: 0.0049
Epoch 96/100
3s - loss: 0.0053 - val_loss: 0.0049
Epoch 97/100
2s - loss: 0.0053 - val_loss: 0.0049
Epoch 98/100
2s - loss: 0.0053 - val_loss: 0.0049
Epoch 99/100
3s - loss: 0.0054 - val_loss: 0.0049
Epoch 100/100
3s - loss: 0.0054 - val_loss: 0.0049
In [49]:
from keras.models import load_model

model = load_model('my_model_RMSprop.h5')

Step 7: Visualize the Loss and Test Predictions

(IMPLEMENTATION) Answer a few questions and visualize the loss

Question 1: Outline the steps you took to get to your final neural network architecture and your reasoning at each step.

Answer: Inspired by the CIFAR10 CNN we've learnt in the previous lessons and the starting architecture from the provided blog post, I first build an CNN with the architecture: 2 Conv2D, 1 MaxPooling, 1 Dropout, 1 Conv2D, 1 MaxPooling, 1 Conv2D, 1 MaxPooling, 1 Dropout, 1 Flatten, 2 Dense, 1 Dropout and 1 Dense. However, the model ends with nearly 10 million parameters, which is relatively large for our dataset (only thousands of 96 * 96 pictures). Thus I add another Conv2D layer, and a MaxPooling layer with larger pool size, which results in about 2 million parameters. The loss of modified architecure decrease faster than the first one, and gives both training and validation loss smaller than 0.0015.

Question 2: Defend your choice of optimizer. Which optimizers did you test, and how did you determine which worked best?

Answer: Three kinds of optimizers are tested: RMSprop, Adadelta and SGD. As we can see, in the test only RMSprop optimizer reaches < 0.0015 loss within 100 epoches. Even with larger learning rates, Adadelta and SGD both converge slower. From the loss curves below, it seems that SGD stuch at some local minimum quickly and improve slowly afterwards, while RMSprop and Adadelta are better at jump around. So in summary, at least for the proposed architecture, RMSprop works better.

Use the code cell below to plot the training and validation loss of your neural network. You may find this resource useful.

In [45]:
## TODO: Visualize the training and validation loss of your neural network
def viz_loss(history, title=None):
    plt.plot(history.history['loss'])
    plt.plot(history.history['val_loss'])
    plt.title(title or 'model loss')
    plt.ylabel('loss')
    plt.xlabel('epoch')
    plt.legend(['train', 'validation'], loc='upper right')
    plt.show()

viz_loss(hist_RMSprop, 'RMSprop')
viz_loss(hist_Adadelta, 'Adadelta')
viz_loss(hist_SGD, 'SGD')

Question 3: Do you notice any evidence of overfitting or underfitting in the above plot? If so, what steps have you taken to improve your model? Note that slight overfitting or underfitting will not hurt your chances of a successful submission, as long as you have attempted some solutions towards improving your model (such as regularization, dropout, increased/decreased number of layers, etc).

Answer: From the plots above, there is hardly any sign of overfitting (i.e. validation loss first decreases and then increases) or underfitting (validation loss is much larger than training loss). Both the training loss and validation loss keeps decreasing, and they become almost the same after 100 epoches.

Visualize a Subset of the Test Predictions

Execute the code cell below to visualize your model's predicted keypoints on a subset of the testing images.

In [50]:
y_test = model.predict(X_test)
fig = plt.figure(figsize=(20,20))
fig.subplots_adjust(left=0, right=1, bottom=0, top=1, hspace=0.05, wspace=0.05)
for i in range(9):
    ax = fig.add_subplot(3, 3, i + 1, xticks=[], yticks=[])
    plot_data(X_test[i], y_test[i], ax)

Step 8: Complete the pipeline

With the work you did in Sections 1 and 2 of this notebook, along with your freshly trained facial keypoint detector, you can now complete the full pipeline. That is given a color image containing a person or persons you can now

  • Detect the faces in this image automatically using OpenCV
  • Predict the facial keypoints in each face detected in the image
  • Paint predicted keypoints on each face detected

In this Subsection you will do just this!

(IMPLEMENTATION) Facial Keypoints Detector

Use the OpenCV face detection functionality you built in previous Sections to expand the functionality of your keypoints detector to color images with arbitrary size. Your function should perform the following steps

  1. Accept a color image.
  2. Convert the image to grayscale.
  3. Detect and crop the face contained in the image.
  4. Locate the facial keypoints in the cropped image.
  5. Overlay the facial keypoints in the original (color, uncropped) image.

Note: step 4 can be the trickiest because remember your convolutional network is only trained to detect facial keypoints in $96 \times 96$ grayscale images where each pixel was normalized to lie in the interval $[0,1]$, and remember that each facial keypoint was normalized during training to the interval $[-1,1]$. This means - practically speaking - to paint detected keypoints onto a test face you need to perform this same pre-processing to your candidate face - that is after detecting it you should resize it to $96 \times 96$ and normalize its values before feeding it into your facial keypoint detector. To be shown correctly on the original image the output keypoints from your detector then need to be shifted and re-normalized from the interval $[-1,1]$ to the width and height of your detected face.

When complete you should be able to produce example images like the one below

In [52]:
# Load in color image for face detection
image = cv2.imread('images/obamas4.jpg')


# Convert the image to RGB colorspace
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
image_copy = np.copy(image)

# plot our image
fig = plt.figure(figsize = (9,9))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])
ax1.set_title('image copy')
ax1.imshow(image_copy)
Out[52]:
<matplotlib.image.AxesImage at 0x7f3035a2d080>
In [67]:
### TODO: Use the face detection code we saw in Section 1 with your trained conv-net 
gray = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)
faces = face_cascade.detectMultiScale(gray, 1.23, 6)

pt_x = np.array([])
pt_y = np.array([])
for (x,y,w,h) in faces:
    cv2.rectangle(image_copy, (x,y), (x+w,y+h), (255,0,0), 3)
    face = cv2.resize(gray[y:y+h, x:x+w], (96,96)) / 255.
    landmarks = np.squeeze(model.predict(np.expand_dims(np.expand_dims(face, axis=-1), axis=0)))
    pt_x = np.hstack((pt_x, ((landmarks[0::2] * 48 + 48)*w/96)+x))
    pt_y = np.hstack((pt_y, ((landmarks[1::2] * 48 + 48)*h/96)+y))
    

## TODO : Paint the predicted keypoints on the test image
fig = plt.figure(figsize = (9,9))
ax = fig.add_subplot(111)
ax.set_xticks([])
ax.set_yticks([])
ax.set_title('image with feature points')
ax.scatter(pt_x, pt_y, marker='o', c='#00FF00', s=30)
ax.imshow(image_copy)
Out[67]:
<matplotlib.image.AxesImage at 0x7f30b56f6898>

(Optional) Further Directions - add a filter using facial keypoints to your laptop camera

Now you can add facial keypoint detection to your laptop camera - as illustrated in the gif below.

The next Python cell contains the basic laptop video camera function used in the previous optional video exercises. Combine it with the functionality you developed for keypoint detection and marking in the previous exercise and you should be good to go!

In [ ]:
import cv2
import time 
from keras.models import load_model
def laptop_camera_go():
    # Create instance of video capturer
    cv2.namedWindow("face detection activated")
    vc = cv2.VideoCapture(0)

    # Try to get the first frame
    if vc.isOpened(): 
        rval, frame = vc.read()
    else:
        rval = False
    
    # keep video stream open
    while rval:
        # plot image from camera with detections marked
        cv2.imshow("face detection activated", frame)
        
        # exit functionality - press any key to exit laptop video
        key = cv2.waitKey(20)
        if key > 0: # exit by pressing any key
            # destroy windows
            cv2.destroyAllWindows()
            
            # hack from stack overflow for making sure window closes on osx --> https://stackoverflow.com/questions/6116564/destroywindow-does-not-close-window-on-mac-using-python-and-opencv
            for i in range (1,5):
                cv2.waitKey(1)
            return
        
        # read next frame
        time.sleep(0.05)             # control framerate for computation - default 20 frames per sec
        rval, frame = vc.read()  
In [ ]:
# Run your keypoint face painter
laptop_camera_go()

(Optional) Further Directions - add a filter using facial keypoints

Using your freshly minted facial keypoint detector pipeline you can now do things like add fun filters to a person's face automatically. In this optional exercise you can play around with adding sunglasses automatically to each individual's face in an image as shown in a demonstration image below.

To produce this effect an image of a pair of sunglasses shown in the Python cell below.

In [ ]:
# Load in sunglasses image - note the usage of the special option
# cv2.IMREAD_UNCHANGED, this option is used because the sunglasses 
# image has a 4th channel that allows us to control how transparent each pixel in the image is
sunglasses = cv2.imread("images/sunglasses_4.png", cv2.IMREAD_UNCHANGED)

# Plot the image
fig = plt.figure(figsize = (6,6))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])
ax1.imshow(sunglasses)
ax1.axis('off');

This image is placed over each individual's face using the detected eye points to determine the location of the sunglasses, and eyebrow points to determine the size that the sunglasses should be for each person (one could also use the nose point to determine this).

Notice that this image actually has 4 channels, not just 3.

In [ ]:
# Print out the shape of the sunglasses image
print ('The sunglasses image has shape: ' + str(np.shape(sunglasses)))

It has the usual red, blue, and green channels any color image has, with the 4th channel representing the transparency level of each pixel in the image. Here's how the transparency channel works: the lower the value, the more transparent the pixel will become. The lower bound (completely transparent) is zero here, so any pixels set to 0 will not be seen.

This is how we can place this image of sunglasses on someone's face and still see the area around of their face where the sunglasses lie - because these pixels in the sunglasses image have been made completely transparent.

Lets check out the alpha channel of our sunglasses image in the next Python cell. Note because many of the pixels near the boundary are transparent we'll need to explicitly print out non-zero values if we want to see them.

In [ ]:
# Print out the sunglasses transparency (alpha) channel
alpha_channel = sunglasses[:,:,3]
print ('the alpha channel here looks like')
print (alpha_channel)

# Just to double check that there are indeed non-zero values
# Let's find and print out every value greater than zero
values = np.where(alpha_channel != 0)
print ('\n the non-zero values of the alpha channel look like')
print (values)

This means that when we place this sunglasses image on top of another image, we can use the transparency channel as a filter to tell us which pixels to overlay on a new image (only the non-transparent ones with values greater than zero).

One last thing: it's helpful to understand which keypoint belongs to the eyes, mouth, etc. So, in the image below, we also display the index of each facial keypoint directly on the image so that you can tell which keypoints are for the eyes, eyebrows, etc.

With this information, you're well on your way to completing this filtering task! See if you can place the sunglasses automatically on the individuals in the image loaded in / shown in the next Python cell.

In [ ]:
# Load in color image for face detection
image = cv2.imread('images/obamas4.jpg')

# Convert the image to RGB colorspace
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)


# Plot the image
fig = plt.figure(figsize = (8,8))
ax1 = fig.add_subplot(111)
ax1.set_xticks([])
ax1.set_yticks([])
ax1.set_title('Original Image')
ax1.imshow(image)
In [ ]:
## (Optional) TODO: Use the face detection code we saw in Section 1 with your trained conv-net to put
## sunglasses on the individuals in our test image

(Optional) Further Directions - add a filter using facial keypoints to your laptop camera

Now you can add the sunglasses filter to your laptop camera - as illustrated in the gif below.

The next Python cell contains the basic laptop video camera function used in the previous optional video exercises. Combine it with the functionality you developed for adding sunglasses to someone's face in the previous optional exercise and you should be good to go!

In [ ]:
import cv2
import time 
from keras.models import load_model
import numpy as np

def laptop_camera_go():
    # Create instance of video capturer
    cv2.namedWindow("face detection activated")
    vc = cv2.VideoCapture(0)

    # try to get the first frame
    if vc.isOpened(): 
        rval, frame = vc.read()
    else:
        rval = False
    
    # Keep video stream open
    while rval:
        # Plot image from camera with detections marked
        cv2.imshow("face detection activated", frame)
        
        # Exit functionality - press any key to exit laptop video
        key = cv2.waitKey(20)
        if key > 0: # exit by pressing any key
            # Destroy windows 
            cv2.destroyAllWindows()
            
            for i in range (1,5):
                cv2.waitKey(1)
            return
        
        # Read next frame
        time.sleep(0.05)             # control framerate for computation - default 20 frames per sec
        rval, frame = vc.read()    
        
In [ ]:
# Load facial landmark detector model
model = load_model('my_model.h5')

# Run sunglasses painter
laptop_camera_go()